diff options
| author | Eli Friedman <eli.friedman@gmail.com> | 2009-06-07 06:52:44 +0000 |
|---|---|---|
| committer | Eli Friedman <eli.friedman@gmail.com> | 2009-06-07 06:52:44 +0000 |
| commit | 3234587213b624e90581208794a17ebfab575a33 (patch) | |
| tree | f9e495f17b9ba8594a6c6d3e80beea42e55e7be5 /llvm/test | |
| parent | cc1305899874dbc4349314684b75235ed25309a2 (diff) | |
| download | bcm5719-llvm-3234587213b624e90581208794a17ebfab575a33.tar.gz bcm5719-llvm-3234587213b624e90581208794a17ebfab575a33.zip | |
Slightly generalize the code that handles shuffles of consecutive loads
on x86 to handle more cases. Fix a bug in said code that would cause it
to read past the end of an object. Rewrite the code in
SelectionDAGLegalize::ExpandBUILD_VECTOR to be a bit more general.
Remove PerformBuildVectorCombine, which is no longer necessary with
these changes. In addition to simplifying the code, with this change,
we can now catch a few more cases of consecutive loads.
llvm-svn: 73012
Diffstat (limited to 'llvm/test')
| -rw-r--r-- | llvm/test/CodeGen/X86/vec_loadsingles.ll | 12 | ||||
| -rw-r--r-- | llvm/test/CodeGen/X86/vec_set-5.ll | 3 | ||||
| -rw-r--r-- | llvm/test/CodeGen/X86/vec_set-6.ll | 2 |
3 files changed, 14 insertions, 3 deletions
diff --git a/llvm/test/CodeGen/X86/vec_loadsingles.ll b/llvm/test/CodeGen/X86/vec_loadsingles.ll new file mode 100644 index 00000000000..67122763ec9 --- /dev/null +++ b/llvm/test/CodeGen/X86/vec_loadsingles.ll @@ -0,0 +1,12 @@ +; RUN: llvm-as < %s | llc -march=x86 -mattr=+sse2 | grep movq + +define <4 x float> @a(<4 x float> %a, float* nocapture %p) nounwind readonly { +entry: + %tmp1 = load float* %p + %vecins = insertelement <4 x float> undef, float %tmp1, i32 0 + %add.ptr = getelementptr float* %p, i32 1 + %tmp5 = load float* %add.ptr + %vecins7 = insertelement <4 x float> %vecins, float %tmp5, i32 1 + ret <4 x float> %vecins7 +} + diff --git a/llvm/test/CodeGen/X86/vec_set-5.ll b/llvm/test/CodeGen/X86/vec_set-5.ll index 4fc652c022a..d3329701119 100644 --- a/llvm/test/CodeGen/X86/vec_set-5.ll +++ b/llvm/test/CodeGen/X86/vec_set-5.ll @@ -1,7 +1,6 @@ ; RUN: llvm-as < %s | llc -march=x86 -mattr=+sse2 -o %t -f ; RUN: grep movlhps %t | count 1 -; RUN: grep movq %t | count 1 -; RUN: grep movsd %t | count 1 +; RUN: grep movq %t | count 2 define <4 x float> @test1(float %a, float %b) nounwind { %tmp = insertelement <4 x float> zeroinitializer, float %a, i32 0 ; <<4 x float>> [#uses=1] diff --git a/llvm/test/CodeGen/X86/vec_set-6.ll b/llvm/test/CodeGen/X86/vec_set-6.ll index 02df526cee8..c7b6747a86f 100644 --- a/llvm/test/CodeGen/X86/vec_set-6.ll +++ b/llvm/test/CodeGen/X86/vec_set-6.ll @@ -1,6 +1,6 @@ ; RUN: llvm-as < %s | llc -march=x86 -mattr=+sse2 -o %t -f ; RUN: grep movss %t | count 1 -; RUN: grep movups %t | count 1 +; RUN: grep movq %t | count 1 ; RUN: grep shufps %t | count 1 define <4 x float> @test(float %a, float %b, float %c) nounwind { |
